In the next decade, energy conservation and greenhouse gas emission reduction will be key issues for developing new vehicles and powertrains.
As is well known, CO2 emissions are straightly related to fuel consumption, and their measurement is carried out on the chassis dynamometer test bench simulating the vehicle road load. However, the measurement of vehicle fuel consumption by the NEDC procedure on the dynamometer test bench is affected not only by experimental uncertainties but also by test tolerances due to vehicle and powertrain parameters as well as to test facility and procedure.
As a consequence, the test-result repeatability (results of different tests with the same vehicle on the same bench) and reproducibility (results of different tests with the same vehicle on different benches)must be carefully taken into account. An overall uncertainty of up to ± 10% can be expected for CO2 emission and fuel consumption figures. Hence, vehicle fuel consumption statistical distributions are more reliable and representative than a single deterministic value.
A new refined Matlab/Simulink model was developed and applied. For each simulation run, vehicle, test and procedure variables were randomly derived from probability density functions with a mean value equal to the nominal variable value, so that a representative statistical distribution can be determined for the vehicle fuel consumption.
Furthermore, different driving cycles were analyzed, for evaluating the variability of vehicle use by the customers.